Finite Element Analysis (FEA) is a method used predominantly in cases where it is hard to predict the behavior of engineering systems using conventional methodology. If it is impossible, or extremely difficult, to create a model based on physical principles, then the next easiest way to tackle it is by conducting experiments, creating graphs from the data and curve-fitting those to get equations that can be used to predict the behavior of the system (empirical equations). A third alternative, or, in some cases, the only possible solution, is to use FEA.
The basic concept of FEA has its roots in a field called ‘numerical methods.’ As shown in the animation, the concept is to divide a surface into many tiny elements, of which the size and location relative to each other are known, and then, starting from the boundary values, to move through all the elements and calculate the properties at any given point in the surface. This is best implemented using Linear Algebra principles, whereby a large matrix (in some cases 10,000 x 10,000 elements big) is calculated using the location, sizes and parameters of the elements and can then be used in a simpler matrix equation to solve for the unknown variables.
FEA is usually used to calculate properties such as temperatures or stresses across structures using commercially available software, but other properties such as forces, coefficients of friction etc can be evaluated as well.